- Email: [email protected]
Interactions of retrotransposons with the host genome: the case of the gypsy element of Drosophila
H~EVIEWS
T r a n s p o s a b l e elements may insert into genes and cause mutant phenotypes by a variety of mechanisms, the nature of which will depend on the location of the insertion site with respect to different structural and functional domains of the affected gene. In addition, the nature of the element itself is important in determining the basis of the phenotype, since transposons carry signals that are not only necessary for various aspects of the transposon's expression, but can also act on adjacent genes or interfere with the transcriptional machinery of these genesL In many cases, the phenotype of transposable element-induced mutations can be suppressed or enhanced by second site mutations at various modifier loci 2q. The genetic behavior of suppressor and enhancer loci, and the specificity of their effect on various types of mutations, may indicate to what extent the nature of the element, rather than its location within the gene, contributes to the phenotype of specific alleles. For example, mutations in some modifier loci, such as suppressor of sable, reverse the phenotype of alleles induced both by the insertion of the 412 retrotransposon5 and by the P element (P. Geyer et al., unpublished). Because of the very different structural and functional characteristics of these two elements, this pattern of interactions suggests that the basis for the mutant phenotype is probably related to some aspect of gene expression affected by the insertion of foreign sequences. In contrast, mutations in the su(Hw)locus exert a more selective effect on transposon-induced mutations, since in general only the phenotypes of alleles resulting from the insertion of the gypsy element are affected by mutations in this gene. This specificity suggests that the mechanism of gypsy mutagenesis is dependent on some idiosyncratic
~ u | |
3' "~ ....
l
II
II
|l|
,"
AILI~
~'~"
5'
----I-
Interactions of retr0transp0s0ns with the host gen0me: the case of the gypsy element of Drosophila VICTOR G. CORCESAND PAMELAK. GEYER Insertion of the gypsy re|rotraasposon into various Drosophila genes resuUs ia mutant phenotypes that can be altered by second site mutations in a variety of modifier loci, One of these loci is the suppressor of Hairy-wing~ which encodes a DNA-binding protein that binds to specific sequences of the gypsy elemeat to regulate its expressio~ Interactions betweea the su(Hw) protein and trattscription factors respoJtsible for expression of the mutant genes are the primary cause of gypsy-imluced phenotypes. Gypsy also appears to mediate effects in trans between copies of a gene located on homologous chromosomes. This interchromosomal communication allows transcriptional enhancers located in one chromosome to interact with their target promoter located on the other bomolog,
properties of this element rather than on the processes affected in the expression of the mutant gene. An understanding of the mechanisms by which insertions of transposable elements result in mutant phenotypes can therefore benefit from the study of the nature of the products encoded by modifier loci and their interaction with specific elements. In addition, the study of these genes, which are identified from their effect on transposable elementinduced mutations, allows the characterizI P u t a t i v e zeste b i n d i n g site ation of important cellular functions for which direct genetic selection is not avail0 su(Hw) protein able. Here we will describe recent advances in our understanding of the variety of effects 3' the gypsy retrotransposon can mediate both on adjacent genes and on sequences located wing blade on the homologous chromosome. body cuticle bristles denticle belts aristae mouth parts denticle belts tarsal claws
FIGg The structure of the yellowlocus, with two exons represented by black boxes separated by an intron. Various transcriptional enhancers involved in the expression of yellowin different tissues are represented by boxes adjacent to the names of the corresponding tissues; the horizontal lines indicate the extent of the sequences previously characterized as containing the various enhancers6, whereas the black squares indicate the most likely locations of these sequences based on further studies (P. Geyer and V. Corces, unpublished). The gypsy element in the y2 allele is inserted at -700 bp from the start of transcription. Filled boxes represent the long terminal repeats and arrows indicate the direction of transcription. Thick vertical lines represent putative zoste-binding sites as deduced from sequence homologies to the published target site for the zeste protein29. The su(Hw)-binding site is indicated by a black circle.
Gypsy-induced mutagenesis at theyellow locus A large part of our understanding of the mechanisms by which the gypsy element mutates adjacent genes comes from studies carried out with the yellow locus. The Xlinked yellow gene encodes a 1.9 kb RNA expressed in late embryo/early larva, and later during the mid-pupal stages of development. Pupal expression of yellow is responsible for the proper coloration of adult structures such as bristles, hairs, wing blades and the thoracic and abdominal cuticle. This pattern of temporal and spatial expression of yellow is controlled by a series of tissue-specific transcriptional enhancers that independently regulate yellow expression in different tissues and
TIG MARCH1991 VOL. 7 NO. 3 ~ Ic,,K)I Els~.'~icr Scient t" Publishers Ltd ( i K ) 0108 - 9479,91,$0200
m
H~EVIEWS
671
AAT.ATTC
ATTC, TA*
T
* T< TAC TTATC y z~J
T GCT< TACI TA j/z~t
ct ~wo
V
V
GAGAAACCAAATAATT TI1TATTGCATAC~TTTTTAATAAAATmATTC-~kTA~,CT CTTTTAATAAAAA~TAT TGCATA~TG bxd TM
ACCAAACAAATT TT~TT(~ATAC~AATAAAAGATTATT~T TGCATA(~G~ T A A T ~ T ~ ~ T A ~ T C T ~ TA HwB$
,
/~K;Iti
structure of the su(Hw)-binding site, showing the sequences located in the 5' transcribed untranslated region of gypsy that interact with the su(Hw) protein. The fragment represented is necessary and sufficient to confer the same phenotype as the complete gypsy element when inserted in the 5' region of the yellowgene. Repeats homologous to the octamer motif are boxed. Also shown are the sites at which insertion of other transposable elements into gypsy-induced alleles results in reversion of the mutant phenotype. Sequences deleted in a bitboraxoidpartial revertant are shown between brackets. stages of development 6. The arrangement of these regulatory elements is summarized in Fig. 1. Insertion of the gypsy element into the yellow gene to create the y2 allele results in a developmentally and spatially restricted phenotype. The mouth parts and denticle belts of the larva as well as the bristles of the adult are wild type, whereas the adult wings and body cuticle are mutant. Gypsy insertion in the y2 allele has taken place at -700 bp from the transcription start site, such that the regulatory sequences that control yellow expression in the wings and body cuticle are now separated from the .yellow promoter by the 7.4 kb gypsy element6. 7 (Fig. 1).
sequences are necessary for the induction of mutant phenotypes (P. Smith and V. Corces, unpublished). Furthermore, flies transformed with a yellow gene containing only these sequences in the same location at which the gypsy element is found in y2 show a y2 phenotype, indicating that the 12 copies of the octamer-like repeat are sufficient to generate the mutant phenotype (P. Geyer and V. Corces, unpublished). The mutagenic effect of these sequences is directional: only enhancers located distal to the insertion site with respect to the promoter are affected.
Specific gypsy sequences are involved in mutagenesis
Null mutations in the su(Hw)gene show a femalesterile phenotype characterized by degeneration of the nurse cells surrounding the developing oocyte. The chromosomes of the mutant nurse cells remain condensed, whereas the normal polytene chromosomes of wild-type nurse cells uncoil before vitellogenesis. The su(Hw) gene encodes a 110 kDa protein that is present at all stages of development and in most tissues of the fly. The amino-terminal region contains a 48 amino acid acidic domain that contains 50% Asp and Glu residues. The central portion of the protein includes 12 copies of the zinc finger motif, and the carboxyterminal portion has several regions of homology with the leucine zipper motif as well as a second acidic domain 13. These three classes of motifs have been found in various types of transcription factors in eukaryotes. The zinc finger domain has been shown to be involved in DNA binding 14, whereas acidic and leucine zipper domains might be involved in protein-protein interactions between transcription factors 15.1~,. The su(Hw) protein is localized in the nucleus and is present at 100--200 different sites on polytene chromosomes from third instar larvae 17. These results agree with a putative role for su(Hw) as a general transcrip tion factor. Biochemical analysis shows that su(Hw~ protein interacts specifically with gypsy sequences. binding to the same octamer-like motif that is necessary and sufficient to confer the characteristic gypsy-induced phenotype (Fig. 2). Adjacent Nr-rich sequences provide a DNA bend on each side of the repeat that is necessary for high-affinity interaction 18.
One obvious explanation for the mutant phenotype of y2 flies is that transcriptional enhancers that regulate .yellow transcription in the two mutant strucu4res are now incapable of acting on the yellow pr6~noter because they are 7.4 kb further away from the TATA box. Analysis of revertants of gypsy-induced mutations indicates that this is not the case8,9. In particular, one class of y2 revertants arose by insertion of jockey or hobo elements into the same region of an intact gypsy element, suggesting that gypsy sequences where insertion has taken place might be involved in the generation of mutant phenotypes9. This region contains 12 copies of a repeated sequence homologous to the octamer motif found in transcriptional enhancers and promoters of eukaryotic genes (see, for example, Ref. 10). Revertants of other gypsy-induced mutations in ttairy-win~) and cut11 are also due to the insertion of other transposable elements in the octamer-like region, and revertants of gypsy-induced bithoraxoid alleles contain deletions of some of the copies of the octamer-like repeats 12 (Fig. 2). More direct evidence for the involvement of these sequences in gypsy mutagenesis was obtained by P-element-mediated transformation of the y2 locus containing the yellow gene plus the gypsy element inserted in the 5' region. When this DNA fragment, containing an intact gypsy element, is injected into y - flies, the transformants show a y2 phenotype. But when the 12 copies of the octamerlike repeat are deleted from the gypsy element, the resulting transformed flies are y+, suggesting that these
Interaction with the product
TIG ,MARCH1991 VOL.7 NO. 3
B
suppressor of Hairy-wing gene
[~EVIEWS These results provide a further indication that su(Hw) protein may act as a transcription factor involved in some general aspect of the regulation of cellular genes. In fact, mutations at the su(Hw)locus result in up to 25-fold reduction in the accumulation of gypsy RNA, suggesting in addition a role for this protein in the control of gypsy expressionT, 19.
The mechanism of ds-acting effects on yellow The mechanism by which the gypsy element causes a mutant phenotype in the y2 allele can be rationalized based on the following observations: (1) the su(Hw) gene encodes a DNA-binding protein containing acidic and leucine zipper domains that may mediate interactions with other proteins; (2) the su(Hw) protein binds to specific sequences in the gypsy element; (3) the su(Hw)-binding site is necessary and sufficient to induce the same mutant phenotype as the intact gypsy element; and (4) the mutagenic effect of these sequences is directional. The conclusion from these observations is that the gypsy element per se is not responsible for the mutant phenotype in the y2 allele, and that gypsy is simply a mediator of the effect of the su(Hw) protein. The reversion of gypsy-induced phenotypes by mutations in the su(Hw)gene can then be explained by the absence in these strains of a functional su(Hw) protein that can bind to the gypsy element. The su(Hw) protein could exert this effect through its acidic and leucine zipper domains, which may interact with other proteins, particularly the tissue-specific transcription factors responsible for the temporal and spatial pattern of yellow expression. Binding of su(Hw) to gypsy sequences may interfere with the action of enhancer-bound factors on the yellow promoter (Fig. 3). The directionality of this effect can be intuitively
explained if the su(Hw) protein interacts with these factors and blocks their effect on transcription initiation. These results agree with models that view enhancer sequences as entry points for specific transcription factors that can then track along the DNA until they encounter the promoter. Because the presence of the su(Hw) protein should not affect DNA looping the effect of this protein is at odds with models that propose looping of the DNA to allow the interaction of enhancer-bound transcription factors with transcription complexes sitting on the promoter of the gene.
Gypsy-induced mutagenesis in other Drosophila genes
The y2 allele represents a particular case in which the gypsy element is inserted into the 5' region of the yellow gene. A similar case is that of bithoraxoid mutations, which are due to the insertion of gypsy into the upstream regulatory region of the Ultrabithorax gene a0 and are probably caused by a mechanism analogous to 3# (Fig. 3). However, other gypsyinduced alleles suppressible by su(Hw)are the result of gypsy insertion within the RNA-coding region. Whatever the cellular function of the su(Hw) protein, the mutant phenotypes caused by gypsy are a direct consequence of the binding of su(Hw) protein to this element, since the lack of su(Hw) protein results in a reversion of the mutant phenotype. The question then arises as to the mechanism of gypsy-mediated mutagenesis in these other alleles. In two cases described, insertion of gypsy into an intron or an exon of a gene results in premature termination of transcription. In the dominant Hairywing mutation Hw~, gypsy insertion causes termination of transcription of the achaete RNA in the gypsy 5' long terminal repeat (LTR), giving rise to a truncated transcript that encodes a functional protein missing the carboxy terminus 2t. The dominant mutant phenotype of this allele is due to ectopic and increased expression of the truncated achaete RNA in the wing imaginal discs ax, • ° °.°°~P"°Ir" • o•o Q° perhaps caused by inter• oo" o#° ,.°° 6° ,~° action of su(Hw) protein • • o,° oO" with regulatory factors • ~ ,~" controlling the expression • ,,J oo°" ~l°,°• ii I|~°°" i1. °l° of the achaete gene. As ..° • ~olmuju, expected from this FIGR! hypothesis, mutations in Mechanism of mutagenesis by the su(Hw) protein. The diagram represents a model of the y2 the su(Hw)gene restore allele, but could be extended to explain other cases discussed in the text. DNA corresponding to normal levels of the trunthe 5' region of the yellowgene is depicted, with RNA polymerase and other TATA-binding cated achaete transcript proteins located in the promoter. The direction of transcription is indicated by an arrow. but do not affect the Transcription factors responsible for yellowexpression in the wing and body cuticle are located premature termination of distally with respect to the promoter. The su(Hw) protein is bound to gypsy sequences transcription 21. Insertion of located between the promoter and the wing and body enhancers. Several functional domains of the su(Hw) protein are indicated: the zinc fingers (Zn) interact with gypsy DNA and the acidic gypsy into an intron of the (Ac) and/or leucine zipper (Leu) domains may interact with transcription factors located hsp82 gene also results in altered termination of tranupstream. This effect is indicated by ~olid arrows, whereas the inability of the wing and body transcription factors to act on the promoter as a consequence of the presence of su(Hw) is scription; in this case, the indicated by dotted lines. presence of the su(Hw) •
o o°
TIG MARCH1991 VOL. 7 NO. 3
[~EVIEWS
B
A XGO
R
"J.LI
X
X H
HH P
I
II
NI
XGO
XGO
R
I
XGO
R
x
x H
I
II
I I I I ! ! I
/
I
II
HH P
XGO
I
wildtype phenotype
)
X H
I y2 XGO
l1
i
G
I
',
R
'J,,l, I
A"LI.
R
X
x H
HH',P
~. I
I
II
',l.J
I y~gb
a~
t
y2
/
phenotype
)__
ye9
/
I "%,,°°.°**
FIG~ Interactions between paired yellow genes located on different chromosomes. Each panel shows the structure of two yellow alleles located on different homologs. (A) A complementing combination of y2 and y59b; (B) the noncomplementing combination of y2 and 3*59. Filled boxes depict yellow exons and the empty box represents the intron. Boxes in the 5' region delineate the wing and body cuticle transcriptional enhancers. Gypsy sequences are shown inserted in the 5' region of yellow; brackets in the y59b and y69 diagrams represent the'breakpoints of the deletion that originated both mutations. The su(Hw) protein is represented by a black circle. Solid arrows indicate the negative effect of su(Hw) on transcription factors bound to tissue-specific enhancers; a broken line indicates the positive effect of enhancers on a promoter in trans, whereas the dotted line indicates the inability' of these enhancers to act on the promoter in trans. target site on the gypsy element potentiates the use of the polyadenylation site located upstream in the 5' LTR, although absence of the protein in su(Hw) mutants does not result in an increase of readthrough transcript levels 23. In two additional cases, insertion of gypsy into the RNA-coding region does not result in transcription termination. Gypsyqnduced bithorax alleles originate from the insertion into different sequences of the third intron of the Ubx gene. The phenotype caused by these insertions increases in intensity as the insertion site approaches the Ubx promoter region; for a constant insertion site, the phenotype is more severe when gypsy insertion takes place in the same transcriptional orientation as the Ubx gene, that is, when the su(Hw)-binding site is closer to the promoter of this gene 24. These results would be difficult to explain if the su(Hw) protein acts primarily on transcription termination, since truncation of the RNA in different regions of the intron would not have different phenotypic effects. A more likely possibility is that the Ubx intron contains different regulatory regions and, as is the case for yellow, only the function of those located distal to the gypsy insertion site with respect to the Ubx promoter would be affected by the presence of this element. Similar reasoning would explain the effect of gypsy insertion into an intron of the forked locus, since the presence of gypsy does not cause truncation of the RNA but rather lower than normal accumulation of wild-type transcripts 25. This effect of gypsy on the rate of transcription initiation could be c o m p o u n d e d with secondary effects on transcription termination, RNA stability, and the ability to splice out
transposable element sequences when gypsy is inserted into an intron of the gene. Interactions between chromosomes
genes on homologous
Understanding the mechanism by which gypsy causes the y2 phenotype enables a series of genetic interactions between y2 and various yellow mutations to be interpreted. Some yellow null alleles give rise to a .~' phenotype when c o m p o u n d e d with y2, whereas other null mutations, such as y59b, give rise to wild-type ye/yWb females. This ability of the 3 ~59b mutation to complement y is dependent on chromosomal pairing between the two yellow alleles, in that disruption of synapsis by translocation of the 3 ~' locus results in a noncomplementing y2 phenotype. In addition, this complementation is affected by mutations at the zeste locus: in the presence of a zeste null allele such as 2 ¢'77h, the phenotype of 3a/y 5°h flies is the same as p-'. Interallelic complementation that is dependent upon pairing and the allelic state of zeste has been termed transvection 26, and therefore the phenomenon described above represents a description of transvection at the yellow locus 27. On the basis of the structure of complementing and noncomplementing yellow alleles, the molecular mechanism of transvection at yellow can be interpreted as the result of the activation of the 3a promoter by enhancers located on the homologous chromosome in the y59b locus27 ('Fig. 4A). This hypothesis is supported by the behavior of alleles in which the enhancers are not functional, resulting in inability to complement .pc (Fig. 4B). Partial inactivation of the enhancers
~G MARCH1991 VOt. 7 NO. 3
m
~EVIEWS results in an intermediate behavior and only partial complementation. The requirements for transvection at the yellow locus are more complex than the simple presence of active enhancers in one of the genes and an intact transcription unit in the other, since several yellow mutations examined follow these requirements but are unable to complement y2. In these noncomplementing alleles, the promoter and enhancer sequences responsible for expression of yellow in the wing and body cuticle are functional, and the mutations affect processes downstream from the transcription initiation step, such as RNA stability or protein structure. These results suggest that the presence of an active promoter in cis may be responsible for the inability of these mutations to complement y2. In agreement with this hypothesis, alleles caused by deletion of promoter sequences or insertions of transposable elements into the promoter region can complement y Z This suggests that the wing and body cuticle enhancers might act preferentially on the yellow promoter located in cis, and that the cis action prevents effects in trans on the promoter located on the homologous chromosome. This ability of enhancers to act in trans has been demonstrated in vitro z8, and offers important clues to the mechanisms normally employed by these sequences to activate transcription from a distant promoter. In all pair-wise combinations of alleles tested, at least one of the pair must contain the gypsy element in order to obtain a complementing wild-type phenotype, raising the question of the putative involvement of gypsy in the transvection phenomenon. A role for gypsy in mediating interchromosomal interactions between the two copies of the yellow gene is supported by the fact that an additional copy of the gypsy element at the scute locus, located close to the yellow gene, precludes transvection effects: yesc1/y59b flies display a 3,2 phenotype, suggesting that gypsy-gypsy interactions in cis interfere with interallelic complementation in trans. The role of gypsy in transvection at yellow may be to mediate effects of the zeste protein, since binding sites for this protein 29 cannot be found in the yellow gene but several clusters of the zeste recognition sequence exist in the gypsy element.
Conclusions To explain the mechanisms of mutagenesis by the gypsy element in spontaneous alleles, as well as the bases for the reversion of the mutant phenotypes by mutations in various modifier loci, we propose a double action of gypsy on the expression of the mutant gene. First, gypsy causes an effect on the rate of transcription initiation of adjacent genes as a consequence of the interaction of the gypsy-bound su(Hw) protein with transcription factors necessary for the expression of the gene (Fig. 3); this effect could result in a decrease or increase in RNA levels depending on whether the transcription factors act as positive or negative regulators of gene expression, respectively. Furthermore, when gypsy insertion takes place in an intron of the gene, the effect on transcription rate is compounded with an effect on the~tability of the precursor transposable element-mutant gene hybrid RNA. This double effect would explain the response of
alleles caused by insertions of gypsy in introns to modifiers such as suppressor o f sable, suppressor o f white-apricot and suppressor o f forked 2, some of which are presumed to act at the level of splicing and RNA stability3,5, since the removal of the gypsy-containing intron is necessary for the formation of a stable message. In addition to these cfs effects, gypsy appears to mediate trans effects on sequences located on the homologous chromosome, adding one more level of complexity to the repertoire of interactions between transposable elements and the genome of the host.
Acknowledgements This work was supported by Public Health Service Award GM35463 and American Cancer Society Grant NP-546A.
References 1 Boeke, J.D. and Corces, V.G. (1989) Annu. Rev. Microbiol. 43, 403-434 2 Rutledge, B.J. etal. (1988) Genetics 119, 391-397 3 Levis,R., O'Hare, K. and Rubin, G.M. (1984) Ce1138, 471--481 4 Modolell, J., Bender, W. and Meselson, M. (1983) Proc. Natl Acad. Sci. USA 80, 1678-1682 5 Fridell, R.A., Pret, A-M. and Searles, L.L. (1990) Genes Dev. 4, 55%566 6 Geyer, P.K. and Corces, V.G. (198,7) GenesDev. 1,996-1004 7 Parkhurst, S.M and Corces, V.G. (1986) Mol. Cell. Biol. 6, 47-53 8 Geyer, P.K., Green, M.M. and Corces, V.G. (1988) Proc. Natl Acad. Sci USA 85, 3938-3942 9 Geyer, P.K., Green, M.M. and Corces, V.G. (1988) Proc. Natl Acad. Sci. USA 85, 8593--8597 10 Rosales, R. et al. (1987) ~ B O J . 6, 301%3025 11 Mizrokhi, L.J. et al. (1985) EMBOJ. 4, 3781-3787 12 Peifer, M. and Bender, W. (1988) Proc. Natlacad. Sci. USA 85, 9650-9654 13 Parkhurst, S.M. etal. (1988) GenesDev. 2, 1205-1215 14 Klug, A. and Rhodes, D. (1987) TrendsBiochem. Sci. 12, 464-469 15 Hope, I.A., Mahadevan, S. and Struhl, K. (1988) Nature 333, 635--640 16 Landschulz, W.H., Johnson, P.F. and McKnight, S.L. (1988) Science 240, 175%1764 17 Spana, C., Harrison, D.A. and Corces, V.G. (1988) Genes Dev. 2, 1414-1423 18 Spana, C. and Corces, V.G. (1990) GenesDev. 4, 150%1515 19 Mazo, AM. et al. (1989) EMBOJ. 8, 903-911 20 Bender, W., Weiffenbach, B., Karch, E and Peifer, M. (1985) Cold Spring Harbor Syrup. Quant. Biol. 50,173-180 21 Campuzano, S. etal. (1986) Cell44, 303-312 22 Balcells, L., Modolell, J. and Ruiz-Gomez, M. (1988) EMBOJ. 7, 389%3906 23 Dorsett, D., Viglianti, G.A., Rutledge, B.J. and Meselson, M. (1989) GenesDev. 3, 454-468 24 Peifer, M. and Bender, W. (1986) EMBOJ. 5, 2293-2303 25 Parkhurst, S.M. and Corces, V.G. (1985) Cell41, 429--437 26 Lewis, E.B. (1954) Am. Nat. 88, 22%239 27 Geyer, P.K., Green, M.M. and Corces, V.G. (1990) EMBO J. 9, 2247-2256 28 Miiller, H-P., Sogo, J.M. and Schaffner, W. (1989) Cell 58, 767-777 29 Benson, M. and Pirrotta, V. (1988) EMBOJ. 7, 3907-3915 I V.G. CORCES AND P.K. GEYER ARE IN THE DEPARTMENT OF BIOLOGY, THE JOHNS HOPKINS UNIVERSITY, BALTIMORE, M D 21218, USA; P.ICG. IS CURRENTLYAT THE DEPARTMENT OF BIOCHEMISTRY, UNIVERSITY OF IOWA COLLEGE OF MEDICINE, IOWA CIT~, IX 52242, USA.
TIG MARCH1991 VOL.7 NO. 3
m
T r a n s p o s a b l e elements may insert into genes and cause mutant phenotypes by a variety of mechanisms, the nature of which will depend on the location of the insertion site with respect to different structural and functional domains of the affected gene. In addition, the nature of the element itself is important in determining the basis of the phenotype, since transposons carry signals that are not only necessary for various aspects of the transposon's expression, but can also act on adjacent genes or interfere with the transcriptional machinery of these genesL In many cases, the phenotype of transposable element-induced mutations can be suppressed or enhanced by second site mutations at various modifier loci 2q. The genetic behavior of suppressor and enhancer loci, and the specificity of their effect on various types of mutations, may indicate to what extent the nature of the element, rather than its location within the gene, contributes to the phenotype of specific alleles. For example, mutations in some modifier loci, such as suppressor of sable, reverse the phenotype of alleles induced both by the insertion of the 412 retrotransposon5 and by the P element (P. Geyer et al., unpublished). Because of the very different structural and functional characteristics of these two elements, this pattern of interactions suggests that the basis for the mutant phenotype is probably related to some aspect of gene expression affected by the insertion of foreign sequences. In contrast, mutations in the su(Hw)locus exert a more selective effect on transposon-induced mutations, since in general only the phenotypes of alleles resulting from the insertion of the gypsy element are affected by mutations in this gene. This specificity suggests that the mechanism of gypsy mutagenesis is dependent on some idiosyncratic
~ u | |
3' "~ ....
l
II
II
|l|
,"
AILI~
~'~"
5'
----I-
Interactions of retr0transp0s0ns with the host gen0me: the case of the gypsy element of Drosophila VICTOR G. CORCESAND PAMELAK. GEYER Insertion of the gypsy re|rotraasposon into various Drosophila genes resuUs ia mutant phenotypes that can be altered by second site mutations in a variety of modifier loci, One of these loci is the suppressor of Hairy-wing~ which encodes a DNA-binding protein that binds to specific sequences of the gypsy elemeat to regulate its expressio~ Interactions betweea the su(Hw) protein and trattscription factors respoJtsible for expression of the mutant genes are the primary cause of gypsy-imluced phenotypes. Gypsy also appears to mediate effects in trans between copies of a gene located on homologous chromosomes. This interchromosomal communication allows transcriptional enhancers located in one chromosome to interact with their target promoter located on the other bomolog,
properties of this element rather than on the processes affected in the expression of the mutant gene. An understanding of the mechanisms by which insertions of transposable elements result in mutant phenotypes can therefore benefit from the study of the nature of the products encoded by modifier loci and their interaction with specific elements. In addition, the study of these genes, which are identified from their effect on transposable elementinduced mutations, allows the characterizI P u t a t i v e zeste b i n d i n g site ation of important cellular functions for which direct genetic selection is not avail0 su(Hw) protein able. Here we will describe recent advances in our understanding of the variety of effects 3' the gypsy retrotransposon can mediate both on adjacent genes and on sequences located wing blade on the homologous chromosome. body cuticle bristles denticle belts aristae mouth parts denticle belts tarsal claws
FIGg The structure of the yellowlocus, with two exons represented by black boxes separated by an intron. Various transcriptional enhancers involved in the expression of yellowin different tissues are represented by boxes adjacent to the names of the corresponding tissues; the horizontal lines indicate the extent of the sequences previously characterized as containing the various enhancers6, whereas the black squares indicate the most likely locations of these sequences based on further studies (P. Geyer and V. Corces, unpublished). The gypsy element in the y2 allele is inserted at -700 bp from the start of transcription. Filled boxes represent the long terminal repeats and arrows indicate the direction of transcription. Thick vertical lines represent putative zoste-binding sites as deduced from sequence homologies to the published target site for the zeste protein29. The su(Hw)-binding site is indicated by a black circle.
Gypsy-induced mutagenesis at theyellow locus A large part of our understanding of the mechanisms by which the gypsy element mutates adjacent genes comes from studies carried out with the yellow locus. The Xlinked yellow gene encodes a 1.9 kb RNA expressed in late embryo/early larva, and later during the mid-pupal stages of development. Pupal expression of yellow is responsible for the proper coloration of adult structures such as bristles, hairs, wing blades and the thoracic and abdominal cuticle. This pattern of temporal and spatial expression of yellow is controlled by a series of tissue-specific transcriptional enhancers that independently regulate yellow expression in different tissues and
TIG MARCH1991 VOL. 7 NO. 3 ~ Ic,,K)I Els~.'~icr Scient t" Publishers Ltd ( i K ) 0108 - 9479,91,$0200
m
H~EVIEWS
671
AAT.ATTC
ATTC, TA*
T
* T< TAC TTATC y z~J
T GCT< TACI TA j/z~t
ct ~wo
V
V
GAGAAACCAAATAATT TI1TATTGCATAC~TTTTTAATAAAATmATTC-~kTA~,CT CTTTTAATAAAAA~TAT TGCATA~TG bxd TM
ACCAAACAAATT TT~TT(~ATAC~AATAAAAGATTATT~T TGCATA(~G~ T A A T ~ T ~ ~ T A ~ T C T ~ TA HwB$
,
/~K;Iti
structure of the su(Hw)-binding site, showing the sequences located in the 5' transcribed untranslated region of gypsy that interact with the su(Hw) protein. The fragment represented is necessary and sufficient to confer the same phenotype as the complete gypsy element when inserted in the 5' region of the yellowgene. Repeats homologous to the octamer motif are boxed. Also shown are the sites at which insertion of other transposable elements into gypsy-induced alleles results in reversion of the mutant phenotype. Sequences deleted in a bitboraxoidpartial revertant are shown between brackets. stages of development 6. The arrangement of these regulatory elements is summarized in Fig. 1. Insertion of the gypsy element into the yellow gene to create the y2 allele results in a developmentally and spatially restricted phenotype. The mouth parts and denticle belts of the larva as well as the bristles of the adult are wild type, whereas the adult wings and body cuticle are mutant. Gypsy insertion in the y2 allele has taken place at -700 bp from the transcription start site, such that the regulatory sequences that control yellow expression in the wings and body cuticle are now separated from the .yellow promoter by the 7.4 kb gypsy element6. 7 (Fig. 1).
sequences are necessary for the induction of mutant phenotypes (P. Smith and V. Corces, unpublished). Furthermore, flies transformed with a yellow gene containing only these sequences in the same location at which the gypsy element is found in y2 show a y2 phenotype, indicating that the 12 copies of the octamer-like repeat are sufficient to generate the mutant phenotype (P. Geyer and V. Corces, unpublished). The mutagenic effect of these sequences is directional: only enhancers located distal to the insertion site with respect to the promoter are affected.
Specific gypsy sequences are involved in mutagenesis
Null mutations in the su(Hw)gene show a femalesterile phenotype characterized by degeneration of the nurse cells surrounding the developing oocyte. The chromosomes of the mutant nurse cells remain condensed, whereas the normal polytene chromosomes of wild-type nurse cells uncoil before vitellogenesis. The su(Hw) gene encodes a 110 kDa protein that is present at all stages of development and in most tissues of the fly. The amino-terminal region contains a 48 amino acid acidic domain that contains 50% Asp and Glu residues. The central portion of the protein includes 12 copies of the zinc finger motif, and the carboxyterminal portion has several regions of homology with the leucine zipper motif as well as a second acidic domain 13. These three classes of motifs have been found in various types of transcription factors in eukaryotes. The zinc finger domain has been shown to be involved in DNA binding 14, whereas acidic and leucine zipper domains might be involved in protein-protein interactions between transcription factors 15.1~,. The su(Hw) protein is localized in the nucleus and is present at 100--200 different sites on polytene chromosomes from third instar larvae 17. These results agree with a putative role for su(Hw) as a general transcrip tion factor. Biochemical analysis shows that su(Hw~ protein interacts specifically with gypsy sequences. binding to the same octamer-like motif that is necessary and sufficient to confer the characteristic gypsy-induced phenotype (Fig. 2). Adjacent Nr-rich sequences provide a DNA bend on each side of the repeat that is necessary for high-affinity interaction 18.
One obvious explanation for the mutant phenotype of y2 flies is that transcriptional enhancers that regulate .yellow transcription in the two mutant strucu4res are now incapable of acting on the yellow pr6~noter because they are 7.4 kb further away from the TATA box. Analysis of revertants of gypsy-induced mutations indicates that this is not the case8,9. In particular, one class of y2 revertants arose by insertion of jockey or hobo elements into the same region of an intact gypsy element, suggesting that gypsy sequences where insertion has taken place might be involved in the generation of mutant phenotypes9. This region contains 12 copies of a repeated sequence homologous to the octamer motif found in transcriptional enhancers and promoters of eukaryotic genes (see, for example, Ref. 10). Revertants of other gypsy-induced mutations in ttairy-win~) and cut11 are also due to the insertion of other transposable elements in the octamer-like region, and revertants of gypsy-induced bithoraxoid alleles contain deletions of some of the copies of the octamer-like repeats 12 (Fig. 2). More direct evidence for the involvement of these sequences in gypsy mutagenesis was obtained by P-element-mediated transformation of the y2 locus containing the yellow gene plus the gypsy element inserted in the 5' region. When this DNA fragment, containing an intact gypsy element, is injected into y - flies, the transformants show a y2 phenotype. But when the 12 copies of the octamerlike repeat are deleted from the gypsy element, the resulting transformed flies are y+, suggesting that these
Interaction with the product
TIG ,MARCH1991 VOL.7 NO. 3
B
suppressor of Hairy-wing gene
[~EVIEWS These results provide a further indication that su(Hw) protein may act as a transcription factor involved in some general aspect of the regulation of cellular genes. In fact, mutations at the su(Hw)locus result in up to 25-fold reduction in the accumulation of gypsy RNA, suggesting in addition a role for this protein in the control of gypsy expressionT, 19.
The mechanism of ds-acting effects on yellow The mechanism by which the gypsy element causes a mutant phenotype in the y2 allele can be rationalized based on the following observations: (1) the su(Hw) gene encodes a DNA-binding protein containing acidic and leucine zipper domains that may mediate interactions with other proteins; (2) the su(Hw) protein binds to specific sequences in the gypsy element; (3) the su(Hw)-binding site is necessary and sufficient to induce the same mutant phenotype as the intact gypsy element; and (4) the mutagenic effect of these sequences is directional. The conclusion from these observations is that the gypsy element per se is not responsible for the mutant phenotype in the y2 allele, and that gypsy is simply a mediator of the effect of the su(Hw) protein. The reversion of gypsy-induced phenotypes by mutations in the su(Hw)gene can then be explained by the absence in these strains of a functional su(Hw) protein that can bind to the gypsy element. The su(Hw) protein could exert this effect through its acidic and leucine zipper domains, which may interact with other proteins, particularly the tissue-specific transcription factors responsible for the temporal and spatial pattern of yellow expression. Binding of su(Hw) to gypsy sequences may interfere with the action of enhancer-bound factors on the yellow promoter (Fig. 3). The directionality of this effect can be intuitively
explained if the su(Hw) protein interacts with these factors and blocks their effect on transcription initiation. These results agree with models that view enhancer sequences as entry points for specific transcription factors that can then track along the DNA until they encounter the promoter. Because the presence of the su(Hw) protein should not affect DNA looping the effect of this protein is at odds with models that propose looping of the DNA to allow the interaction of enhancer-bound transcription factors with transcription complexes sitting on the promoter of the gene.
Gypsy-induced mutagenesis in other Drosophila genes
The y2 allele represents a particular case in which the gypsy element is inserted into the 5' region of the yellow gene. A similar case is that of bithoraxoid mutations, which are due to the insertion of gypsy into the upstream regulatory region of the Ultrabithorax gene a0 and are probably caused by a mechanism analogous to 3# (Fig. 3). However, other gypsyinduced alleles suppressible by su(Hw)are the result of gypsy insertion within the RNA-coding region. Whatever the cellular function of the su(Hw) protein, the mutant phenotypes caused by gypsy are a direct consequence of the binding of su(Hw) protein to this element, since the lack of su(Hw) protein results in a reversion of the mutant phenotype. The question then arises as to the mechanism of gypsy-mediated mutagenesis in these other alleles. In two cases described, insertion of gypsy into an intron or an exon of a gene results in premature termination of transcription. In the dominant Hairywing mutation Hw~, gypsy insertion causes termination of transcription of the achaete RNA in the gypsy 5' long terminal repeat (LTR), giving rise to a truncated transcript that encodes a functional protein missing the carboxy terminus 2t. The dominant mutant phenotype of this allele is due to ectopic and increased expression of the truncated achaete RNA in the wing imaginal discs ax, • ° °.°°~P"°Ir" • o•o Q° perhaps caused by inter• oo" o#° ,.°° 6° ,~° action of su(Hw) protein • • o,° oO" with regulatory factors • ~ ,~" controlling the expression • ,,J oo°" ~l°,°• ii I|~°°" i1. °l° of the achaete gene. As ..° • ~olmuju, expected from this FIGR! hypothesis, mutations in Mechanism of mutagenesis by the su(Hw) protein. The diagram represents a model of the y2 the su(Hw)gene restore allele, but could be extended to explain other cases discussed in the text. DNA corresponding to normal levels of the trunthe 5' region of the yellowgene is depicted, with RNA polymerase and other TATA-binding cated achaete transcript proteins located in the promoter. The direction of transcription is indicated by an arrow. but do not affect the Transcription factors responsible for yellowexpression in the wing and body cuticle are located premature termination of distally with respect to the promoter. The su(Hw) protein is bound to gypsy sequences transcription 21. Insertion of located between the promoter and the wing and body enhancers. Several functional domains of the su(Hw) protein are indicated: the zinc fingers (Zn) interact with gypsy DNA and the acidic gypsy into an intron of the (Ac) and/or leucine zipper (Leu) domains may interact with transcription factors located hsp82 gene also results in altered termination of tranupstream. This effect is indicated by ~olid arrows, whereas the inability of the wing and body transcription factors to act on the promoter as a consequence of the presence of su(Hw) is scription; in this case, the indicated by dotted lines. presence of the su(Hw) •
o o°
TIG MARCH1991 VOL. 7 NO. 3
[~EVIEWS
B
A XGO
R
"J.LI
X
X H
HH P
I
II
NI
XGO
XGO
R
I
XGO
R
x
x H
I
II
I I I I ! ! I
/
I
II
HH P
XGO
I
wildtype phenotype
)
X H
I y2 XGO
l1
i
G
I
',
R
'J,,l, I
A"LI.
R
X
x H
HH',P
~. I
I
II
',l.J
I y~gb
a~
t
y2
/
phenotype
)__
ye9
/
I "%,,°°.°**
FIG~ Interactions between paired yellow genes located on different chromosomes. Each panel shows the structure of two yellow alleles located on different homologs. (A) A complementing combination of y2 and y59b; (B) the noncomplementing combination of y2 and 3*59. Filled boxes depict yellow exons and the empty box represents the intron. Boxes in the 5' region delineate the wing and body cuticle transcriptional enhancers. Gypsy sequences are shown inserted in the 5' region of yellow; brackets in the y59b and y69 diagrams represent the'breakpoints of the deletion that originated both mutations. The su(Hw) protein is represented by a black circle. Solid arrows indicate the negative effect of su(Hw) on transcription factors bound to tissue-specific enhancers; a broken line indicates the positive effect of enhancers on a promoter in trans, whereas the dotted line indicates the inability' of these enhancers to act on the promoter in trans. target site on the gypsy element potentiates the use of the polyadenylation site located upstream in the 5' LTR, although absence of the protein in su(Hw) mutants does not result in an increase of readthrough transcript levels 23. In two additional cases, insertion of gypsy into the RNA-coding region does not result in transcription termination. Gypsyqnduced bithorax alleles originate from the insertion into different sequences of the third intron of the Ubx gene. The phenotype caused by these insertions increases in intensity as the insertion site approaches the Ubx promoter region; for a constant insertion site, the phenotype is more severe when gypsy insertion takes place in the same transcriptional orientation as the Ubx gene, that is, when the su(Hw)-binding site is closer to the promoter of this gene 24. These results would be difficult to explain if the su(Hw) protein acts primarily on transcription termination, since truncation of the RNA in different regions of the intron would not have different phenotypic effects. A more likely possibility is that the Ubx intron contains different regulatory regions and, as is the case for yellow, only the function of those located distal to the gypsy insertion site with respect to the Ubx promoter would be affected by the presence of this element. Similar reasoning would explain the effect of gypsy insertion into an intron of the forked locus, since the presence of gypsy does not cause truncation of the RNA but rather lower than normal accumulation of wild-type transcripts 25. This effect of gypsy on the rate of transcription initiation could be c o m p o u n d e d with secondary effects on transcription termination, RNA stability, and the ability to splice out
transposable element sequences when gypsy is inserted into an intron of the gene. Interactions between chromosomes
genes on homologous
Understanding the mechanism by which gypsy causes the y2 phenotype enables a series of genetic interactions between y2 and various yellow mutations to be interpreted. Some yellow null alleles give rise to a .~' phenotype when c o m p o u n d e d with y2, whereas other null mutations, such as y59b, give rise to wild-type ye/yWb females. This ability of the 3 ~59b mutation to complement y is dependent on chromosomal pairing between the two yellow alleles, in that disruption of synapsis by translocation of the 3 ~' locus results in a noncomplementing y2 phenotype. In addition, this complementation is affected by mutations at the zeste locus: in the presence of a zeste null allele such as 2 ¢'77h, the phenotype of 3a/y 5°h flies is the same as p-'. Interallelic complementation that is dependent upon pairing and the allelic state of zeste has been termed transvection 26, and therefore the phenomenon described above represents a description of transvection at the yellow locus 27. On the basis of the structure of complementing and noncomplementing yellow alleles, the molecular mechanism of transvection at yellow can be interpreted as the result of the activation of the 3a promoter by enhancers located on the homologous chromosome in the y59b locus27 ('Fig. 4A). This hypothesis is supported by the behavior of alleles in which the enhancers are not functional, resulting in inability to complement .pc (Fig. 4B). Partial inactivation of the enhancers
~G MARCH1991 VOt. 7 NO. 3
m
~EVIEWS results in an intermediate behavior and only partial complementation. The requirements for transvection at the yellow locus are more complex than the simple presence of active enhancers in one of the genes and an intact transcription unit in the other, since several yellow mutations examined follow these requirements but are unable to complement y2. In these noncomplementing alleles, the promoter and enhancer sequences responsible for expression of yellow in the wing and body cuticle are functional, and the mutations affect processes downstream from the transcription initiation step, such as RNA stability or protein structure. These results suggest that the presence of an active promoter in cis may be responsible for the inability of these mutations to complement y2. In agreement with this hypothesis, alleles caused by deletion of promoter sequences or insertions of transposable elements into the promoter region can complement y Z This suggests that the wing and body cuticle enhancers might act preferentially on the yellow promoter located in cis, and that the cis action prevents effects in trans on the promoter located on the homologous chromosome. This ability of enhancers to act in trans has been demonstrated in vitro z8, and offers important clues to the mechanisms normally employed by these sequences to activate transcription from a distant promoter. In all pair-wise combinations of alleles tested, at least one of the pair must contain the gypsy element in order to obtain a complementing wild-type phenotype, raising the question of the putative involvement of gypsy in the transvection phenomenon. A role for gypsy in mediating interchromosomal interactions between the two copies of the yellow gene is supported by the fact that an additional copy of the gypsy element at the scute locus, located close to the yellow gene, precludes transvection effects: yesc1/y59b flies display a 3,2 phenotype, suggesting that gypsy-gypsy interactions in cis interfere with interallelic complementation in trans. The role of gypsy in transvection at yellow may be to mediate effects of the zeste protein, since binding sites for this protein 29 cannot be found in the yellow gene but several clusters of the zeste recognition sequence exist in the gypsy element.
Conclusions To explain the mechanisms of mutagenesis by the gypsy element in spontaneous alleles, as well as the bases for the reversion of the mutant phenotypes by mutations in various modifier loci, we propose a double action of gypsy on the expression of the mutant gene. First, gypsy causes an effect on the rate of transcription initiation of adjacent genes as a consequence of the interaction of the gypsy-bound su(Hw) protein with transcription factors necessary for the expression of the gene (Fig. 3); this effect could result in a decrease or increase in RNA levels depending on whether the transcription factors act as positive or negative regulators of gene expression, respectively. Furthermore, when gypsy insertion takes place in an intron of the gene, the effect on transcription rate is compounded with an effect on the~tability of the precursor transposable element-mutant gene hybrid RNA. This double effect would explain the response of
alleles caused by insertions of gypsy in introns to modifiers such as suppressor o f sable, suppressor o f white-apricot and suppressor o f forked 2, some of which are presumed to act at the level of splicing and RNA stability3,5, since the removal of the gypsy-containing intron is necessary for the formation of a stable message. In addition to these cfs effects, gypsy appears to mediate trans effects on sequences located on the homologous chromosome, adding one more level of complexity to the repertoire of interactions between transposable elements and the genome of the host.
Acknowledgements This work was supported by Public Health Service Award GM35463 and American Cancer Society Grant NP-546A.
References 1 Boeke, J.D. and Corces, V.G. (1989) Annu. Rev. Microbiol. 43, 403-434 2 Rutledge, B.J. etal. (1988) Genetics 119, 391-397 3 Levis,R., O'Hare, K. and Rubin, G.M. (1984) Ce1138, 471--481 4 Modolell, J., Bender, W. and Meselson, M. (1983) Proc. Natl Acad. Sci. USA 80, 1678-1682 5 Fridell, R.A., Pret, A-M. and Searles, L.L. (1990) Genes Dev. 4, 55%566 6 Geyer, P.K. and Corces, V.G. (198,7) GenesDev. 1,996-1004 7 Parkhurst, S.M and Corces, V.G. (1986) Mol. Cell. Biol. 6, 47-53 8 Geyer, P.K., Green, M.M. and Corces, V.G. (1988) Proc. Natl Acad. Sci USA 85, 3938-3942 9 Geyer, P.K., Green, M.M. and Corces, V.G. (1988) Proc. Natl Acad. Sci. USA 85, 8593--8597 10 Rosales, R. et al. (1987) ~ B O J . 6, 301%3025 11 Mizrokhi, L.J. et al. (1985) EMBOJ. 4, 3781-3787 12 Peifer, M. and Bender, W. (1988) Proc. Natlacad. Sci. USA 85, 9650-9654 13 Parkhurst, S.M. etal. (1988) GenesDev. 2, 1205-1215 14 Klug, A. and Rhodes, D. (1987) TrendsBiochem. Sci. 12, 464-469 15 Hope, I.A., Mahadevan, S. and Struhl, K. (1988) Nature 333, 635--640 16 Landschulz, W.H., Johnson, P.F. and McKnight, S.L. (1988) Science 240, 175%1764 17 Spana, C., Harrison, D.A. and Corces, V.G. (1988) Genes Dev. 2, 1414-1423 18 Spana, C. and Corces, V.G. (1990) GenesDev. 4, 150%1515 19 Mazo, AM. et al. (1989) EMBOJ. 8, 903-911 20 Bender, W., Weiffenbach, B., Karch, E and Peifer, M. (1985) Cold Spring Harbor Syrup. Quant. Biol. 50,173-180 21 Campuzano, S. etal. (1986) Cell44, 303-312 22 Balcells, L., Modolell, J. and Ruiz-Gomez, M. (1988) EMBOJ. 7, 389%3906 23 Dorsett, D., Viglianti, G.A., Rutledge, B.J. and Meselson, M. (1989) GenesDev. 3, 454-468 24 Peifer, M. and Bender, W. (1986) EMBOJ. 5, 2293-2303 25 Parkhurst, S.M. and Corces, V.G. (1985) Cell41, 429--437 26 Lewis, E.B. (1954) Am. Nat. 88, 22%239 27 Geyer, P.K., Green, M.M. and Corces, V.G. (1990) EMBO J. 9, 2247-2256 28 Miiller, H-P., Sogo, J.M. and Schaffner, W. (1989) Cell 58, 767-777 29 Benson, M. and Pirrotta, V. (1988) EMBOJ. 7, 3907-3915 I V.G. CORCES AND P.K. GEYER ARE IN THE DEPARTMENT OF BIOLOGY, THE JOHNS HOPKINS UNIVERSITY, BALTIMORE, M D 21218, USA; P.ICG. IS CURRENTLYAT THE DEPARTMENT OF BIOCHEMISTRY, UNIVERSITY OF IOWA COLLEGE OF MEDICINE, IOWA CIT~, IX 52242, USA.
TIG MARCH1991 VOL.7 NO. 3
m